基于变量相关性特征聚类的多场景分布式电源规划  

Multi-scenario Distributed Generation Planning Based on Clustering of Variable Correlation Features

作　　者：卢光辉 滕欢[1,2] 廖寒逊 LU Guanghui;TENG Huan;LIAO Hanxun(College of Electrical Engineering,Sichuan University,Chengdu 610065,China;Intelligent Electric Power Grid Key Laboratory of Sichuan Province(Sichuan University),Chengdu 610065,China)

机构地区：[1]四川大学电气工程学院,成都610065 [2]智能电网四川省重点实验室(四川大学),成都610065

出　　处：《电力系统及其自动化学报》2020年第12期95-101,共7页Proceedings of the CSU-EPSA

摘　　要：分布式电源和负荷功率的时序随机性给分布式电源规划带来巨大挑战。首先,将全年各时段的风速、光照强度和负荷大小组成多元时序数据并进行变量相关性特征聚类,得到全年典型规划场景以及对应的权重;然后,建立以分布式电源投资商收益、配电网公司收益最大化为优化目标的分布式电源规划模型,并采用非支配排序遗传算法2对模型进行求解得到帕累托最优解集,采用基于熵权的逼近理想解排序决策法对帕累托最优解集进行多目标折中决策,得到规划方案。仿真算例验证了所提出方法的有效性。The randomness in the time series of distributed generations(DGs)and load power poses a huge challenge to the planning of DGs.In this paper,the multivariate time series data are formed at first by the wind speed,light intensi⁃ty and load size at various time intervals of one year,which are further processed by clustering of variable correlation features,thereby obtaining the typical annual planning scenarios and their corresponding weights.Then,a DG planning model is established with the optimization objective of maximizing the revenues of the DG investor and the distribution network company.In addition,the non dominated sorting genetic algorithm 2(NSGA2)is used to solve the model to ob⁃tain the Pareto optimal solution set,on which multi-objective compromise decisions are made using the entropy weight based TOPSIS decision method.As a result,the planning scheme is obtained.The simulation result of an example veri⁃fies the effectiveness of the proposed method.

关 键 词：多元时序 变量相关性特征聚类 分布式电源投资商收益 配电网公司收益 

分 类 号：TM715[电气工程—电力系统及自动化]